Constructing Peace After Civil War:
Assessing the Role of Power-sharing on the Durability of Peace
by
Marisa L. Martín
MPP Essay
Submitted to
Oregon State University
In partial fulfillment of
the requirements for the
degree of
Master of Public Policy
Presented June 13, 2013
Commencement June 15, 2013
Master of Public Policy essay of Marisa L. Martín presented on June 13, 2013
APPROVED:
Sarah Henderson, representing Political Science
Alison Johnston, representing Political Science
Joseph Orosco, representing Philosophy and Peace Studies
Marisa Martín, Author
ACKNOWLEDGEMENTS
I would like to express my deepest appreciation to my husband and my family for their endless
encouragement and love. I would also like to thank my friends for their constant support. I
extend my profound gratitude to my committee members, Dr. Sarah Henderson, Dr. Alison
Johnston, and Dr. Joseph Orosco, for their support, guidance, and feedback regarding my essay.
It was a great pleasure and an honor to work with such positive, intelligent, and dedicated
individuals. I’d also like to thank Dr. Paul Kopperman for being an outstanding mentor during
my years at Oregon State University. I offer my sincere appreciation to the MPP director, Dr.
Brent Steel, and to all of the instructors and faculty in the MPP program for their contribution to
my education.
AN ABSTRACT OF THE THESIS OF
Marisa L. Martín for the degree of Master of Public Policy presented on June 13, 2013
Title: Constructing Peace After Civil War: Assessing the Role of Power-sharing on the
Durability of Peace
Abstract approved:
Sarah Henderson
A growing concern among the international community is that civil war has become the most
common form of armed conflict worldwide. To prevent further bloodshed, more research is urgently
needed to understand the causes of conflict and most successful strategies for peace. The bargaining
theory of war proposes that former warring parties defect from peace processes as a means of selfpreservation in the face of security threats and uncertainty. But which institutional arrangements are
most effective in mitigating these conditions and facilitating peace? Increasingly, these theorists and
conflict negotiators alike are proposing that power-sharing, arrangements within peace agreements
that guarantee all factions a share of government power, is the answer. Indeed, power-sharing has
become one of the most popular instruments for ending wars today. Nevertheless, its effectiveness is
hotly debated. Consequently, one of the foremost questions within the civil war literature is: does
power-sharing contribute to lasting peace and if so, under what conditions? To answer this question,
this paper makes use of a new data set of 195 peace agreements signed between 1975 and 2008 that
aimed to settle 53 civil conflicts in 44 countries. I find that political power-sharing largely decreases
the risk of war relapse, military power-sharing mitigates conflict under certain conditions, and
territorial power-sharing does not significantly affect peace outcomes. Thus, power-sharing is an
important tool for ending war, however, other arrangements, such as disarmament, demobilization,
and reintegration of soldiers, amnesty, elections, and comprehensive agreements are more critical to
establishing long-term peace.
TABLE OF CONTENTS
Page
INTRODUCTION………………………………………………………………………………...1
LITERATURE REVIEW…………………………………………………………………………2
DATA AND RESEARCH METHODS…………………………………………………………26
RESULTS………………………………………………………………………………………..31
DISCUSSION AND POLICY IMPLICATIONS………………………………………………..39
CONCLUSION…………………………………………………………………………………..43
REFERENCES…………………………………………………………………………………..45
LIST OF TABLES & FIGURES
Page
FIGURE A: Power-Sharing Agreements that Resulted in Durable Peace………………………32
TABLE 1: Logistic Regression Results for Influence of Political Power-Sharing on Peace
Outcomes………………………………………………………………………………...34
TABLE 2: Logistic Regression Results for Influence of Military Power-Sharing on Peace
Outcomes……………………………..………………………………………………….36
TABLE 3: Logistic Regression Results for Influence of Territorial Power-Sharing on Peace
Outcomes…………………………..………………………………………………….…38
Introduction
The 21st century has presented a new set of challenges for the global community and has
made policymakers, conflict mediators, and scholars increasingly concerned with how to end
civil war. International changes, such as the spread of globalization and democratization, have
contributed to significant interdependence between countries making interstate war far less
likely, but it has also facilitated the rise of nonstate actors, such as terrorists and insurgents, who
now pose a greater threat to international security (Nye, 2011). In fact, civil war has become the
most common and most deadly form of armed conflict on the planet (Collier, 2007; Gates and
Strom, 2007). From 2010 to 2011, the world experienced the greatest surge of intrastate conflicts
since the 1990s peak of more than 50 civil conflicts (Themnér and Wallensteen, 2012). Of
particular concern is that civil wars tend to last much longer than interstate wars and have proven
more difficult to resolve (Walter 2002). These trends have created a greater sense of urgency
among conflict-resolution practitioners to improve their understanding of the mechanisms that
create lasting peace (Gates and Strom, 2007).
The field of civil war research has grown substantially in the past two decades and
scholars and negotiators have learned many important lessons in that time (Blattman and Miguel,
2010). Yet, post-conflict situations continue to be ripe with unpredictable circumstances and
efforts to establish peace frequently fail (Stedman, 2001). For example, during the 1990s, failed
agreements led to sharp increases in violence in Rwanda, Angola, Somalia, and Sierra Leone.
Why did these peace processes fail? What makes the difference between renewed violence and
enduring peace in nations emerging from conflict? Numerous theories explain that a return to
war is the result of structural or cultural factors; however, the literature is increasingly
recognizing the role that individual interests play. For example, the bargaining theory of war
1 advises that peace breaks down due to conditions of uncertainty and perceived threats among the
actors during the peace process. Bargaining theorists and conflict management practitioners
argue that a successful negotiation process requires institutional arrangements that address these
individual concerns and that provide incentives for cooperation (Norris, 2008). Bargaining
theorists propose that power-sharing is one such arrangement; a perspective that has been largely
adopted by experts in the field. In fact, power-sharing has become one of the most popular
instruments for ending wars today. However, the capacity of power-sharing to stimulate peace
has been highly contested and it is therefore critical to continue exploring whether this widely
used tool actually contributes to lasting peace following civil wars, and if so, under what
conditions. Using quantitative analysis, I find that power-sharing can successfully diminish the
risk of conflict, but that other institutional arrangements, such as disarmament, demobilization,
and reintegration of soldiers, amnesty, elections, and comprehensive agreements are more
important predictors of long-term peace.
Literature Review
Despite great efforts by mediators to end civil conflict, countries that have experienced
civil war are at exceptionally high risk of returning to war, a phenomenon that is often referred to
as a ‘conflict trap’ (Collier, 2007). On average, civil wars last about 5 years; however, “only half
of countries with a conflict that ends manage to make it through a decade without relapse”
(Collier, 2007). Since 2000, ninety percent of all armed conflicts have transpired in nations that
had previously experienced civil war (Walter, 2011). These patterns are disturbing because the
financial and human consequences of intrastate conflict are dire. Economists who have
quantified the average yearly costs of civil war estimate that domestic costs approach $64 billion
and global cost reach $100 billion (Collier, 2007). Wars also result in death, disease, and
2 poverty, and are prime breeding grounds for terrorists and international criminal organizations.
These costs not only devastate the afflicted country, but routinely spillover into other countries.
Undoubtedly, the international community holds a stake in preventing and ending civil wars;
however, achieving and keeping the peace has proven problematic.
Causes of Renewed War
Why do wars recur and how can negotiators stop this cycle? Myriad theories have
emerged to explain the causes of renewed war and the conditions under which lasting peace is
possible. Generally, the factors that cause a society to return to war are the same as those that
cause initial outbreaks of conflict (Walter, 2011). While it is difficult to make definitive
statements regarding causality, the most prominent theories explain renewed conflict as a
consequence of structural factors, cultural factors, dynamics of the previous war, or
characteristics of the peace process (including individual interests, the terms of the settlement,
and third-party involvement) (Walter, 2011; Demmers, 2012).
Structuralist theories emphasize that conflict is path-dependent. This approach explains
how initial country conditions that fail to meet human needs lead to grievances among the
population and motivate rebellion (Demmers, 2012). For example, when political, geopolitical,
and economic rules and structures break down, the opportunity for conflict rises. Thus, the
systems of society, including the political economy, play a major role in igniting conflict.
There are several ways that political and judicial institutions contribute to conflict. For
example, a history of colonialism, authoritarianism, and a corrupt criminal justice system is
associated with greater conflict risk (Demmers, 2012). This is largely because illegitimate and
oppressive regimes alienate populations and motivate insurgency (Stedman, 2001). However,
strong authoritarian governments are often quite effective in blocking rebellions. Thus, the
3 opportunity for rebellion is strongest when political and judicial institutions are too weak to
suppress an uprising (Stedman 2001; Walter, 2011). Additionally, even if the violence comes to
a halt, peace is less likely to last when the regime is corrupt or weak because rebel groups are
less likely to trust the government and because the parties have less experience addressing
problems peacefully (Walter, 2011; Collier, 2007). That is, peace is more durable in democratic
societies with high institutional capacity because actors with prior democratic experience are
more likely to trust one another as well as the democratic characteristics of the peace process
(Hartzell and Hoddie, 2003; Jarstad and Nilsson, 2008).
The characteristics of neighboring states also affect peace outcomes because belligerents
are more likely to return to war if neighboring states offer support (financial, military, or
logistical) to one of the rival parties or if neighboring states express disapproval of the peace
process (Stedman, 2001). The impact of neighboring states is particularly strong when the nation
at war is institutionally weak. Given these political and geopolitical factors, it is commonly
recommended that negotiators improve the nation’s institutional capacity, strengthen government
transparency and accountability, and ensure greater opportunities for public participation and
representation.
Structuralist theories also stress that countries characterized by low per capita income and
poor economic development are prone to cycles of violence (Collier, 2007; Blattman and
Miguel, 2010). Poverty and economic inequality are associated with myriad social problems,
including high mortality, low life expectancy, and a lack of housing, education, healthcare, and
job opportunities. Environments that offer civilians few opportunities often breed resentment
against the government, making it far easier for rebel groups to recruit soldiers (Maiese, 2003;
Walter, 2011). Similar problems occur in countries with a high dependence upon the production
4 and exportation of valuable natural resources such as oil, gas, diamonds, and rare minerals
(Collier, 2007). Production of such resources brings wealth to political elites, which gives them
no reason to change the economic system and incentivizes them to stop investing in human
capital and other sectors of the economy (which in turn exacerbates grievances). Moreover,
government elites and insurgent leaders are often able to use commodity wealth to finance their
war (Stedman, 2001; Collier, 2007; Walter, 2011). Researchers who have focused on economic
explanations of war typically recommend that mediators prioritize economic growth, and remedy
harmful natural resource extraction and trade policies (Maiese, 2003).
Theories grounded in social psychology tend to explain war as the result of unresolved
cultural disputes (Walter, 2011). For example, social identity theory explains that wars sparked
by identity differences, including perceived ethnic, racial, and religious disparities, are
particularly persistent (Hartzell and Hoddie, 2003; Demmers, 2012). According to this theory,
people construct their identities based around a shared set of socially meaningful characteristics
in order to satisfy their need to belong (Demmers, 2012). In doing so, people create in-groups
and out-groups and develop their identity in comparison to others. To protect and reinforce their
sense of identity and self-worth, groups become motivated to improve their social position
relative to other groups, which may embolden oppressive behavior against outsiders. Some
studies indicate that it is easier to recruit and mobilize soldiers based on shared identity than on
shared political or economic ideology because identity issues evoke a deeper emotional response
and greater sense of personal threat (Mason, Brant, Gurses, and Quinn, 2011). It follows that the
number of identity groups in a society affects post-war outcomes. For example, societies with a
small number of sizable ethnic groups are at a higher risk of conflict than societies that are
relatively homogenous or that are extremely fragmented, because large ethnic groups have
5 superior ability to dominate others (Mason, Brant, Gurses, Quinn, 2011; Walter, 2011). The
prevailing wisdom is that resolving identity issues requires helping the groups meet their identity
needs through alternative routes; although policy recommendations for achieving this are not
well developed (Demmers, 2012). One strategy involves helping members of different groups
develop a common identity by emphasizing their shared problems (such as natural disasters or
threats from other nations) or by focusing on their common goals. However, these strategies may
not be realistic in practice and could create other tensions, particularly if focusing on external
threats increases fears. Thus, a more viable alternative may be to encourage members of different
groups to join several new cultural, political, and social groups because “cross-cutting
memberships make the individual less dependent upon one single in-group for the satisfaction of
his or her psychological needs and less vulnerable to polarization” (Demmers, 2012).
Other theorists emphasize that conflict dynamics, or characteristics of the previous war,
impact the odds of future conflict. For example, the length of war matters because parties that
have been involved in conflict longer are more likely to recognize that they are unlikely to ever
win and will therefore have greater incentive to remain at peace (Hartzell and Hoddie, 2003;
Jarstad and Nilsson, 2008). Additionally, wars of high intensity, with high casualty rates, are
associated with elevated risk of renewed conflict because this condition creates more distrust and
resentment (Hartzell and Hoddie, 2003). Lastly, higher numbers of warring groups and soldiers
(e.g. over 50,000 soldiers) introduce greater uncertainty and therefore greater potential that at
least some actors will defect from peace (Stedman, 2001). In short, war is highly path-dependent.
Bargaining Theory of War
While economic, political, or social identity grievances help explain motivations and
opportunities for war, some argue that they are not sufficient to explain why war recurs in some
6 environments with these conditions but not others (Walter, 2011). The bargaining theory of war
(or the rationalist approach) is a dominant theory that emphasizes the strategic nature of war and
argues that individual factors drive conflict recurrence (Lake, 2010). Specifically, asymmetric
information and commitment problems present during the peace process will cause actors to
experience security threats that make them believe that a return to war will lead to better
outcomes than peace (Walter, 2011; Blattman and Miguel, 2010; Mattes and Savun, 2010;
Demmers, 2012).
The theory assumes that warring parties consist of self-interested, utility-maximizing
actors who make rational decisions given perfect information (Blattman and Miguel, 2010).
Bargaining theory’s game-theoretic model assumes that when faced with the options of adhering
to or defecting from peace, players will strategically choose the option with the greatest payoff
(Demmers, 2012). A central idea is that because war is costly, groups with a clear understanding
of the risks should recognize that peace offers the greatest chance for survival (Demmers, 2012).
Following that logic, civil wars that end in a clear military victory should be more likely to result
in long-term cooperation because they leave the defeated group(s) with few resources, little
capacity to recommence conflict, and a clear understanding of the opposition’s greater military
capabilities (Mattes and Savun, 2010). In this case, the victor has no reason to keep fighting and
the losing party has enough information to recognize that committing to peace is the safest
option. Other civil wars end simply because the parties’ resources have been exhausted. In this
situation, both groups have had enough time in battle to learn that they are unlikely to defeat the
other and therefore “both should prefer a bargained solution to destructive conflict” (Blattman
and Miguel, 2010). In short, given the significant risks of fighting and a clear understanding of
7 the likely outcome of war, it would be in the rational interest of these parties to negotiate a
similar outcome peacefully.
Bargaining theorists, however, recognize that warring groups may not always behave
rationally (Blattman and Miguel, 2010). Actors may let their emotions influence their decisions
or may lack access to the information they need to accurately weigh the full costs and benefits of
their decisions. For example, many conflicts come to a close without a clear outcome, such as
those ended via negotiated agreement (either self-enforced or enforced by third parties) (Hartzell
and Hoddie, 2007). Under these circumstances, no decisive victor has been established, meaning
that each party has asymmetric information regarding the relative military capabilities and
intentions of adversaries (Gates and Strom, 2007; Wallensteen, 2001). This uncertainty decreases
trust among the groups and contributes to communication problems. Groups often lie or hide
information out of fear that it could be used against them (Fearon, 1995). For example, they may
overstate their military capabilities in an attempt to appear less vulnerable or to try leverage this
power to get a better deal during the peace process (Fearon, 1995; Mattes and Savun, 2010).
Consequently, adversaries experience commitment problems; meaning parties cannot credibly
promise to one another to uphold the peace (Blattman and Miguel, 2010).
Under such conditions of uncertainty and distrust, groups experience security threats and
become more likely to return to war preemptively as a means of self-preservation (Gates and
Strom, 2007). Hence, rather than being advantageous, overstating one’s military capabilities (or
actually being more powerful) is dangerous because it increases fear among opponents and often
compels them to behave preemptively. However, actors are only likely to succeed in returning to
war if their resources, including weapons, leaders, financing, number of soldiers, and combat
training, are enough to pose a threat to their adversary (Blattman and Miguel, 2010).
8 Accordingly, actors who both oppose or distrust the peacemaking process and have the capacity
to resume conflict are called spoilers (Stedman, 2001; Kieh, 2011). Generally, weaker parties are
more likely to spoil the peace because they experience a greater sense of threat, have less ability
to negotiate a peace deal in their favor, and have a lower opportunity cost of returning to war
(Stedman, 2001; Walter, 2011). Therefore, because insurgents often have less power than the
government and tend to be less satisfied with the status quo, they are more likely to renege on a
peace deal (Gates and Strom, 2007; Mattes and Savun, 2010).
In sum, the rationalist approach explains renewed war as a failure of the bargaining
process and “an inefficient outcome that all parties would avoid in the absence of bargaining
imperfections” (David Lake). Certainly, a poorly designed agreement or one that is implemented
at the wrong moment may make the parties want to return to war (Ouellet, 2004). Nevertheless,
while a clear military victory may provide groups with complete information, much of the
international community agrees that the price of sustained civil war is so high that it is preferable
to initiate a peace process rather than hope that the war will resolve itself (Stedman, 2001).
However, the argument for ending war through a negotiated peace process is not merely a
normative one. Emerging empirical evidence casts doubt on the idea that peace is more durable
following military victory. For example, a recent study by Mason, Gurses, Brandt, and Quinn
(2011) specifies that while peace is often fragile immediately following the signing of a
settlement, when settlements are designed right they are more likely to facilitate long-term peace
than a military victory (Hartzell and Hoddie, 2007; Stedman, 2001). Therefore, given the desire
to avoid a long and devastating battle, the bulk of the literature states that resolving conflicts in
the absence of a clear victory requires getting the terms of the agreement right.
9 Bargaining theorists argue that it is possible to end war in the face of asymmetric
information and commitment problems by reducing security threats and providing parties with
incentives to get them to support the peace process (Stedman, 2001). Without their commitment
to stop fighting, the nation remains vulnerable to another conflict. To facilitate elite buy-in,
bargaining theory makes two main recommendations. First, address the asymmetric information
problem by equalizing groups’ resources. Leveling the playing field should increase transparency
regarding relative capabilities and reduce fear. Second, reduce commitment problems by
rewarding groups for cooperating, while also requiring them to engage in costly concessions
(Jarstad and Nilsson, 2008). The rationalist perspective’s sub-theory of cost-signaling indicates
that during the peace process, parties must be willing to take on ‘costly concessions’ or personal
risks in order to credibly signal that they are serious about committing to peace. In other words,
peace may be constructed by designing agreements that address individual concerns and that
provide warring factions with incentives to commit to peace.
Constructing Peace Through Negotiated Settlements
Parties remain optimistic about their potential to win the war in the absence of a clear
outcome, and therefore entering into an agreement is an important costly behavior that signals
commitment to peace (Stedman, 2001). This process increases trust and encourages groups to
stick to the terms of the settlement. Regrettably, no easy formulas exist for drafting a peace
accord because settlements must provide solutions appropriate to the unique circumstances of
each war. However, there are some broad procedural guidelines as numerous studies have
provided insight into which policies are important to include.
A peace agreement may take the form of an official document or an informal verbal
agreement, public statement, or letter (Yawanarajah and Ouellet, 2003). In some cases, it may be
10 possible to draft a single document. However, peace processes usually begin with the drafting of
several partial agreements over the course of many months or years (which may stand-alone or
be incorporated into a comprehensive agreement at a later date). For example, ceasefire
agreements, in which all parties agree to stop the war for a specified time frame, are often the
first partial agreements to be drafted. Once the fighting has stopped, there are three general
stages to the negotiation process (Hartzell and Hoddie, 2003).
The first stage is pre-negotiation, which outlines the procedural rules and institutions for
negotiating peace, the parties that will be involved in the negotiations and their roles, and where
negotiations will occur (Yawanarajah and Ouellet, 2003). This phase may also include interim or
preliminary agreements wherein former combatants officially agree to negotiate a settlement.
These partial agreements are important for initiating the peace process, but they are fragile as
they typically exclude certain conflict actors and do not fully address the issues at stake
(Yawanarajah and Ouellet, 2003). To enhance the legitimacy of the process, it is best to include
all relevant parties in negotiations and tackle the full range of problems that led to the conflict
(Jarstad, 2008; Brosché, 2005).
Therefore, during the second phase of the peace process, mediators often aim to establish
a comprehensive peace agreement (CPA) (Yawanarajah and Ouellet, 2003). The Uppsala
Conflict Data Program (2013) and the Peace Accords Matrix (2012) define a CPA as a written
document created through the course of negotiation signed by all actors involved in the conflict.
CPAs identify the full range of substantive issues of concern and outline how resources will be
distributed as well as how future disputes will be addressed, including any political, social, legal,
military, or economic structural or procedural changes required (Yawanarajah and Ouellet,
2003).
11 The third phase designates how the settlement will be implemented (Hartzell and Hoddie,
2003). This may involve a separate implementation agreement that specifies actors’
responsibilities, the consequences for not fulfilling their roles, and whether a third party will
oversee the implementation (e.g. the United Nations, the African Union, or an external
government). While settlements facilitate peace, there is a significant margin for error.
Consequently, a pressing objective is to determine which provisions yield the greatest prospects
for long-term conflict termination (Mattes and Savun, 2010).
Given the multitude of theoretical perspectives regarding the main causes of civil
conflict, however, it is understandable that there remains substantial disagreement regarding
which policies to prioritize (Blattman and Miguel, 2010). Despite this disagreement, the
international community generally recognizes disarmament, demobilization, and reintegration of
soldiers (DDR), amnesty, reconciliation, elections, and security guarantees such as powersharing, and third party involvement as the most important terms to include in a settlement (Kieh
Jr., 2011; Stedman, 2001; Mattes and Savun, 2009). For instance, most scholars agree that
because disarmament and demobilization of soldiers make groups vulnerable, these processes
help them signal to one another that they can be trusted to uphold the peace (Stedman, 2001).
Additionally, a failure to help former combatants reintegrate into civilian life, particularly in a
fragile society, makes them more prone to reengage in violence. DDR is presently considered
one of the best predictors of enduring peace and is critical to the initial stages of a peace process
(Jarstad and Nilsson, 2008; Stedman, 2001).
Once soldiers have laid down their arms, it is widely recommended that the nation
address past injustices that contributed to the conflict, including denial of rights, war crimes, and
crimes against humanity (including genocide, torture, forced disappearance, and slavery)
12 (Maiese, 2003). To do so, mediators may include provisions to establish institutions for war
crimes tribunals, war reparations, and criminal justice system reform. A common way to address
past injustices is to create truth and reconciliation commissions, which may take on a variety of
forms. Quite often, nongovernmental organizations, governments, and conflict management
practitioners set up reconciliation commissions (Malek, 2013). For example, South Africa’s
government was instrumental in forming South Africa’s Truth and Reconciliation Commission
that aimed to address the injustices of apartheid, and nongovernmental organizations
spearheaded successful reconciliation commissions in Rwanda following the genocide.
Reconciliation may be more successful, however, when community members and local leaders
initiate these processes. These commissions may resemble formal courts in charge of
investigating and prosecuting war crimes or may be informal without the authority to prosecute.
Generally, commissions are tasked with healing communities by uncovering and publicly
acknowledging the realities of the conflict. Reconciliation can last months or even years as they
call for individuals to come forward one by one to share personal testimony of the war.
Perpetrators are asked to acknowledge their actions and victims are encouraged to forgive them.
Predictably, these processes are long and painful, but because reconciliation opens a dialogue to
address core issues, adversaries that have participated in such processes are more likely to
uphold the terms of the settlement. Many argue that truth and reconciliation commissions are
“the best—or only—way to achieve a full accounting of crimes committed against fellow
citizens and to prevent future conflict” (Rotberg and Thompson, 2010).
It is imperative to note, however, that pursuing justice may be incompatible with the
objective of peace (Stedman, 2001). One of the central difficulties with resolving conflicts is that
warring groups may refuse to put down their arms if it means revealing their role in the war and
13 facing prosecution. Moreover, it may not be feasible to prosecute every perpetrator because
during the course of war, “the line between offenders and victims can become blurred”
(Yawanarajah and Ouellet, 2003). Given these difficulties, settlements may stipulate that parties
acknowledge past injustices (e.g. through reconciliation commissions), but grant parties amnesty
to appease their fears (Yawanarajah and Ouellet, 2003; Rotberg and Thompson, 2010).
Amnesty involves legally exempting perpetrators of crimes from punishment, often with
the condition that parties publically acknowledge guilt during reconciliation processes
(Melander, 2009). Bargaining theory argues that amnesty is supportive of peace because people
have a strong desire for justice and so agreeing to allow one’s enemies to go unpunished is a
costly concession and therefore signals commitment to cessation of conflict. However, there is
debate regarding how effective amnesty is in securing peace. Melander’s (2009) study found that
amnesty might support peace in authoritarian regimes, but not in democracies or in nations
transitioning between regime types. This may be because authoritarian regimes offer a more
stable national environment and because people within authoritarian regimes may be more
accepting of impunity for guilty parties. In contrast, other studies indicate that amnesty may
promote peace in democracies, but only when the most egregious human rights violators and
leaders of the conflict are brought to justice (Snyder and Vinjamuri, 2003). The relationship
between amnesty and peace is thorny to say the least; critics argue that while amnesty may
support peace it does so by sacrificing justice and human rights, and encourages a climate of
impunity for criminal behavior (Melander, 2009). To balance the desire for both peace and
justice, many third parties grant amnesty to the majority of soldiers, but not to the leaders of the
war. Put simply, amnesty allows many guilty parties to avoid legal prosecution and denies justice
to many victims, but may be one of the only ways to get warring elites to agree to stop fighting.
14 Because democratization is associated with peace, many negotiators include elections as
a major provision within settlements in an attempt to help the nation become a stable democracy
(Stedman, 2001). However, democracy requires that elections be free and competitive, involving
candidates who hold legitimacy among constituents. Yet, in post-conflict societies where
political parties are often weak and corrupt, democratic elections may not be feasible.
Consequently, negotiators and scholars are increasingly recognizing that elections, which help
establish how power will be distributed, may be viewed more realistically as an instrument for
ending war than for democratization.
A common recommendation for incentivizing groups to adhere to peace is to convert
warring groups into political parties (Stedman, 2001). These parties may not be perceived as
legitimate, but granting warring elites the chance to compete for political power through the
electoral system is often effective in getting core actors to stop fighting. For example, although
Charles Taylor was considered corrupt and guilty of war crimes, “most international observers
believed that the election of Charles Taylor in 1998 [although undemocratic] was necessary to
bring [Liberia’s] civil war to a close” because he threatened to return to war if he was not elected
(Stedman, 2001). Nevertheless, elections are not always successful in ending war; losing groups
may decide that war will bring more favorable outcomes and there is also significant danger to
putting extremists in power. For example, while Liberia was experiencing short-term peace,
Taylor was actively fueling the brutal civil war in Sierra Leone by providing assistance to the
rebels who were committing egregious human rights violations (Human Rights Watch, 2012).
Furthermore, international observers argue that his oppressive regime led to the outbreak of the
Second Liberian Civil War that occurred about a decade later. Clearly, post-conflict elections are
unlikely to resemble those in fully institutionalized democracies and pose their own set of risks.
15 Nevertheless, several studies have provided evidence that holding elections is successful in
teaching former adversaries how to compete for power through peaceful means and often sets the
stage for consolidating democracy in the country (Stedman, 2001; Jarstad, 2009).
The rationalist approach argues that the key to reducing the risk of war and incentivizing
groups to stick to peace is to provide them with two main security guarantees, power-sharing and
third party involvement (Hartzell and Hoddie, 2003; Norris, 2008). The modern concept of
power-sharing grew out of Arend Lijphart’s theory of consociational democracy, an
institutionalized form of managing conflict in societies divided along ethnic, racial, or religious
lines, in which leaders of warring groups are guaranteed a share of political power (Lijphart,
1969; Norris, 2008). Lijphart argued that holding elections or converting warring groups into
political parties is not enough. He contended that majoritarian democracies (winner-take-all
systems) would prove unstable in societies plagued by conflict because groups that lose the
election will lack a voice in the new government and may be tempted to return to war (Lijphart,
1999). He contended that the only way to mitigate the potential for conflict is to guarantee all
political elites a share of political power (with or without elections). By his definition, political
power-sharing involves four major components: (1) grand coalitions, in which elites from each
rival group are integrated into the executive (often established through the electoral system); (2)
a system of mutual veto power, in which decision-making requires a super-majority or consensus
among all parties in the executive; (3) proportional representation for all groups in civil service
appointments and elections; and (4) self-governing autonomy, in which groups are granted
power to govern over local issues (e.g. through federalism) (Lijphart, 1993; Norris, 2008).
Consociationalism was first developed into a theory during the late 1960s (Norris, 2008).
Over the years, the concept has been widely studied and broadly expanded (Hartzell and Hoddie,
16 2003). Other scholars have coined the terms ‘consensus democracy’, ‘proportional democracy’,
‘negotiation democracy’, and most commonly ‘power-sharing’ to refer to strategies similar to
consociational democracy (Norris, 2008; Jarstad, 2009). Some scholars make a distinction
between dividing resources and sharing resources. For instance, if resources are not shared but
are allocated to groups to control independently, or if certain groups are prohibited from
participating in various government institutions or decision-making processes, many scholars
would label this power dividing rather than power-sharing (Brosché, 2005; Gates and Strom,
2007). Yet a fair number of scholars make no such distinction, and rather stick to a broader
definition of power-sharing.
Hartzell and Hoddie (2003) were principal developers of the broader and most widely
recognized concept of power-sharing. They define contemporary power-sharing to be “those
rules that, in addition to defining how decisions will be made by groups within the polity,
allocate decision-making rights, including access to state resources, among collectivities
competing for power” (Hartzell and Hoddie, 2003). A major contribution of theirs was defining
four distinct forms of power-sharing: political, military, territorial, and economic power-sharing.
They believed that in post-conflict societies, where the concerns of former belligerents are
multidimensional, it is beneficial for political elites to share not only political power, but other
power as well. For example, groups that have been economically marginalized may not agree to
a peace deal unless they are granted economic power. Therefore, the modern definition of powersharing incorporates elements of Lijphart’s consociationalism, but it has evolved.
Political power-sharing involves sharing government power, via mechanisms such as
Lijphart’s grand coalitions, mutual veto power, and proportional representation in government or
civil service (either through elections or appointments) (Walter, 2002; Hartzell and Hoddie,
17 2003; Mattes and Savun, 2009). Territorial power-sharing is where the state continues to have
central control, but local groups are granted greater autonomy over local issues (which can be
especially effective when the crux of the conflict involves geographic disputes) (Mattes and
Savun, 2009; Jarstad and Nilsson, 2008). Federalism is a chief example of territorial powersharing. Military power-sharing involves sharing coercive resources, integrating rebels into the
state security forces, or creating a new national army that guarantees positions to all former
combatants (Mattes and Savun, 2009). Economic power-sharing provides groups mutual control
over state economic resources, and often involves redistributing resources toward historically
marginalized groups (Hartzell and Hoddie, 2003; Mattes and Savun, 2009). In short, powersharing involves a compromise that guarantees all factions a role in government and a share of at
least some political, economic, territorial and/or military power in return for adhering to the
peace (Hartzell and Hoddie, 2003; Brosché, 2005; Jarstad and Nilsson, 2008; Norris, 2008).
While their definition is relatively new, Hartzell and Hoddie (2003) find evidence that these
types of arrangements were implemented in negotiated settlements as early as the mid 1940s in
Colombia, India, and Malaysia.
Effectiveness of Power-sharing
While there are numerous theories regarding the causes and solutions of civil war, there
are important reasons to focus on power-sharing. In recent decades, power-sharing has become
one of the most widely accepted tools for ending civil conflict (particularly among the Western
states). It is one of the most frequently used mechanisms for terminating conflicts, particularly
those in Africa (Mehler 2009; Gates and Strom, 2007). For example, of the 38 civil wars
terminated via negotiated settlement between 1945 and 1999, all but one included power-sharing
provisions (Hartzell and Hoddie, 2003). From 1989 to 2004, 70 of 83 peace agreements
18 contained power-sharing (Brosché, 2005). Since the turn of the 21st century, most negotiated
settlements have included power-sharing provisions. While power-sharing has been extensively
studied and has contributed to peace in several countries, it warrants further examination because
the effectiveness of this widely used tool is highly debated within the literature (Brosché, 2005;
Norris, 2008).
Numerous studies have found evidence that power-sharing provides important benefits.
Among the most frequently cited success stories in which power-sharing has prevented a return
to war for at least a few years are South Africa (1994), Liberia (2003), and Angola (2002)
(Norris, 2008; Jarstad, 2009; Mattes and Savun, 2009). The rationalist approach argues that
power-sharing is effective at reducing the risk of conflict because it equalizes resources and
requires players to make costly concessions (Hartzell and Hoddie, 2003; Norris, 2008).
Providing a share of power to all belligerents makes the competing sides more evenly matched
and allows them to have a clearer understanding of one another’s relative capabilities and
intentions. In turn, security threats and the risk of preemptive war are reduced. Power-sharing is
costly because it restricts the potential power of each party. In other words, all groups must
concede the prospect of gaining full control of the country. Additionally, agreeing to share power
with a former enemy may be perceived as “selling out their interests” and may cause them to
lose supporters (Jarstad and Nilsson, 2008; Mattes and Savun, 2009). Hence, in taking on these
risks, parties are able to credibly signal to one another their intention to commit to peace
(Hartzell and Hoddie, 2003; Gates and Strom, 2007; Jarstad and Nilsson, 2008; Walter, 2002).
Power-sharing is therefore purported to reduce commitment problems, fear, and uncertainty, as
well as encourage cooperation, moderate behavior, and trust (Mattes and Savun, 2010; Brosché,
2005; Norris, 2008; Walter, 2002; Jarstad and Sisk, 2008; Hartzell and Hoddie, 2007).
19 While power-sharing is a costly concession, it also provides each party with a guaranteed
reward—all parties retain at least some power (Walter, 2002). This is important because warring
groups are hesitant to commit to peace if it means surrendering power. Indeed, Walter (2002)
finds that “parties are 38% more likely to sign an accord if it includes a guarantee to be part of
the future government.” According to Gates and Strom (2007), “the temptation to fight is low as
long as the value of power-sharing is greater than the share a spoiler could earn through
fighting.” Thus, by providing a benefit to all warring groups, each group has greater incentive to
remain at peace.
Much of the support for power-sharing comes from large-N studies (Mattes and Savun
2009; Walter 2002; Gates and Strom, 2007). Many of these studies have examined the aggregate
effect of power-sharing dimensions. For example, Hartzell and Hoddie’s (2003) quantitative
study of 38 civil wars between 1945 and 1998 provided evidence that including more dimensions
of power-sharing has a cumulative and positive association with peace. For example, including
all four forms of power-sharing (political, economic, territorial, and military) facilitates peace
better than including only one form of power-sharing. In all of these ways, power-sharing is
considered to support the chances for conflict resolution (Stedman, 1997; Mattes and Savun,
2010; Gates and Strom, 2007).
In contrast, recent studies that have examined power-sharing dimensions independently
raise questions about their effectiveness. For instance, political power-sharing is the most
common form included in peace accords and has been most widely studied (Mehler, 2009;
Walter, 2002). Mattes and Savun (2009) find that “each additional dimension of political powersharing reduces the risk of settlement failure by about 29 percent” and they argue that because it
is the only type of power-sharing that they find to be statistically significant it should be given
20 priority in negotiated settlements. Stedman (2001) also finds evidence that political powersharing significantly raises the probability of peace. Binningsbø’s (2005) study of power-sharing
in 118 post-conflict societies between 1985 and 2002 lends support for political power-sharing as
a valuable instrument for peace, although this study provides for a more nuanced result:
proportional representation may support peace, while grand coalitions do not. To confuse
matters, however, some studies find that political power-sharing undermines peace (although this
has not proven to be statistically significant) or has no effect (Derouen, Lea, and Wallensteen,
2009; Jarstad and Nilsson, 2008; Hartzell and Hoddie, 2003; Mukherjee, 2006).
Territorial power-sharing is less commonly included in settlements (Mehler, 2009).
However, some research indicates that territorial arrangements reduce the risk of relapse
(Binningsbø, 2005; Stedman, 2001). A few studies have implied that military and territorial pacts
are more effective in promoting peace than political or economic pacts because military and
territorial provisions require costlier concessions, thereby signaling greater commitment (Jarstad
and Nilsson, 2008). Yet other studies provide conflicting results. Negotiators have been least
likely to include economic power-sharing in negotiated settlements and therefore few studies
have examined its role on conflict termination. It could be that political power-sharing includes
economic power-sharing, because groups with control over state decisions will also make
decisions regarding economic resources.
Qualitative studies have sparked concern not only over the effectiveness of powersharing, but also its potential dangers. For example, Jarstad (1994) states that a disagreement
over a power-sharing arrangement led to an escalation of violence during the 1994 genocide in
Rwanda. Also worrisome is that power-sharing may actually incentivize groups to initiate war as
a means of obtaining at least some future power (Tull and Mehler, 2004; Sisk, 2003). Power-
21 sharing pacts failed to prevent the outbreak of recommenced war in Lebanon (1975), Northern
Ireland (1974), Somalia (1992), Angola (1992 and 1994), Sierra Leone (1999), and
Czechoslovakia (1993) (Gates and Strom, 2007; Sisk, 2003). The Dayton Agreement following
the Bosnia and Herzegovina conflict is another well-known failure (Norris, 2008).
A major debate over power-sharing is whether it is supportive of democracy (a common
long-term goal among conflict negotiators) and whether it needs to be democratic at all (Gates
and Strom, 2007). Many argue that power-sharing supports democracy because it teaches rivals
basic democratic forms of governance (Lijphart, 1969; Norris, 2008; Gates and Strom, 2007;
Jarstad, 2009; Jarstad and Sisk, 2008; Walter, 2002). However, researchers are increasingly
concerned that it may conflict with the goal of democracy because it is a rigid system that is
inherently elite-centered (Jarstad, 2009). Rather than holding free and competitive elections,
power-sharing guarantees power to groups typically in the absence of elections. If elections are
held, the outcomes are essentially rigged. Therefore, power-sharing may lead to adverse
selection in which it rewards and includes violent extremists who lack popular support while
excluding parties that were not part of the war (Jarstad, 2009; Gates and Strom, 2007; Brosché,
2005). Consequently, power-sharing may reinforce divisions between former belligerents (Sisk,
2003; Tull and Mehler, 2005). Power-sharing may also create a moral hazard by empowering
political elites who may not represent the public’s interests and who have little accountability to
the public (Gates and Strom, 2007). Though elites may be satisfied with power-sharing
arrangements, citizens often are not (Mehler, 2009; Horowitz, 2009). Power-sharing also
involves transaction costs or political stalemates because of its emphasis on inclusion, mutual
veto-power, and consensus-based decision-making in a pluralistic environment (Gates and
Strom, 2007). Notwithstanding, others argue that the very logic of power-sharing relies on the
22 assumption that a majoritarian, winner-take-all, election system following civil war would prove
less stable, and therefore it is necessary to sacrifice democracy and good governance in the shortterm to ensure immediate security (Gates and Strom, 2007).
In practice, power-sharing has been defined in countless ways, had different goals, and
been implemented differently, which has contributed to momentous debate regarding how to
measure, study, and draw conclusions about it (Brosché, 2005; Jarstad, 2009). Unfortunately, the
inconsistencies within the literature are due to differences in methodology, operationalization,
coding, case selection, and interpretation of provisions within peace deals (Mattes and Savun,
2009; Mehler, 2009). This uncertainty has made it more difficult for conflict management
practitioners to design the most effective power-sharing arrangements. Nevertheless, proponents
and critics alike agree that there are few alternatives for ending war, which explains why it is
widely used despite remaining questions (Jarstad, 2009; Brosché, 2005; Mehler, 2009).
Certainly, power-sharing is not perfect, but because it is one of only a handful of tools available
to mediators, it is essential that researchers continue to examine the circumstances under which
power-sharing is effective. Researchers have identified several of these conditions for success.
Conditions for Success
There is disagreement about whether it is acceptable to forgo democracy for peace
through power-sharing in the short-term, but most scholars agree that long-term peace requires a
transition to a more flexible democratic system that allows new coalitions to form that move
beyond the cleavages of war and include previously excluded groups (Jarstad, 2009; Ouellet,
2004; Sisk, 2003). For example, many argue that power-sharing was only successful in the case
of South Africa because the terms of the settlement required the government to transition from
power-sharing institutions towards a democracy within a period of five years (Gates and Strom,
23 2007). It is therefore widely recommended that settlements identify deadlines for phasing out
power-sharing and transitioning toward more permanent systems of democratic elections and
governance (Krienbuehl, 2010).
Hartzell and Hoddie (2003) have found that combatants are often able to reduce their
distrust if substantial costly signals are in place, such as power-sharing and amnesty. However,
even with these provisions, parties may continue to experience informational asymmetries and
commitment problems (Blattman and Miguel, 2010). Hence, most civil war scholars argue that
consolidating peace requires other security guarantees, such as a credible and neutral third party
to monitor and enforce the terms of the settlement (Brosché, 2005; Walter, 2002; Gates and
Strom, 2007; Stedman, 2001, Mattes and Savun, 2009). Hartzell and Hoddie (2003) find that
third parties decrease the risk of a failed agreement by 83%. As a monitor, third parties may
collect, verify, and pass information between the groups, as well as report attempts to violate the
peace deal (Walter, 2002; Mattes and Savun, 2010). For example, former UN Secretary Kofi
Annan and his negotiation team were instrumental in maintaining post-election peace in Kenya
following its civil war. As monitors, the team assisted the groups by identifying the substantive
issues of concern, resolving disputes throughout the peace process, and engendering trust
(Horowitz, 2009).
Third parties are said to be most effective when they have the power to enforce the
settlement by promising to intervene if either party attempts to spoil the peace (Mattes and
Savun, 2010; Kieh Jr., 2011). Multiple reports point to competent third party enforcers as a
major predictor of post-conflict peace (Stedman, 2001; Walter, 2002). Third parties may enforce
the peace through sanctions (e.g. trade sanctions or freezing of assets) or by intervening in a
peacekeeping capacity. The presence of peacekeepers with the mandate to use force is
24 significantly associated with greater chances for long-term peace (Jarstad and Nilsson, 2008;
Mason, Brandt, Gurses, and Quinn, 2011).
Third parties often facilitate conflict resolution; however, some studies have found that
third party intervention has no effect on peace (Fortna, 2004). More worrisome is that third party
intervention, including peacekeeping, has in some cases increased commitment problems and
aggravated violence. A major reason for this failure is that many third parties are willing to
become involved as monitors, but not as enforcers (Stedman, 2001). Being an effective enforcer
requires substantial military capacity and a willingness to endure costs, commit resources, and
risk troops. These are costs that international nongovernmental organizations and external
governments are often unwilling or unable to take on. When third parties merely monitor the
peace process, the warring groups quickly recognize that the monitor lacks the power to protect
them or to punish them for spoiling the peace. When third parties lack the capacity to enforce the
peace, the competing groups have less incentive to stop fighting and may even attack third party
monitors (Blattman and Miguel, 2010). Quite understandably then, many scholars argue that
third parties with a personal stake in ending the conflict will be more effective because they have
greater political will to intervene (Mukherjee, 2006). However, while third parties with a
strategic interest in ending the war may be more effective under some circumstances, they also
run the risk of losing legitimacy among the warring groups. For instance, they may be more
inclined to give preferential treatment to the actors that better serve their interests, or may fail to
properly report or respond to violations of the terms of the agreement. For example, third parties
in Angola did not report certain violations because they were afraid that doing so would make
them seem less neutral, and they falsely reported that the rebel group had demobilized because
they wanted the mission to appear successful (Stedman, 2001). However, other scholars explain
25 that the positive association between intervention and renewed war may simply reflect a spurious
correlation. That is, third parties and peacekeepers are more likely to intervene in the most high
profile and most violent wars that already have the greatest risk of failure. In fact, since the Cold
War, peacekeepers have been more commonly deployed in those situations in which the
resolution of armed conflict was particularly difficult (Fortna, 2004).
Many civil war scholars are beginning to argue that power-sharing is most successful in
creating peace when it is used as a short-term solution, efforts are made to transition toward a
more democratic form of governance, and a third party enforcer oversees the peace process. Yet
questions remain as to why power-sharing has failed to prevent war in many countries,
suggesting that further research is needed to better understand the conditions that make powersharing more or less successful. As Gates and Strom (2007) argue, there are “many reasons to
think that the effects of power-sharing are worth further scrutiny - and that the field could benefit
from new theoretical ideas as well as from improvements in data and methods.”
Data and Research Methods
This paper advances our understanding of the effect of power-sharing on long-term peace
following civil wars. Specifically, this paper tests the hypothesis that political, military, and
territorial power-sharing provisions decrease the probability of renewed violence among the
same parties within five years of signing the peace agreement, when controlling for other factors
that affect post-conflict outcomes including other agreement provisions, conflict dynamics, and
initial conditions of the country (e.g. structural and cultural factors).
To explore the conditions necessary for constructing peace, I utilize the Uppsala Conflict
Data Program’s (UCDP) Peace Agreement Dataset v. 2.0, 1975-2011 that includes data on peace
settlements signed between two warring groups that engaged in armed conflict between 1975-
26 2011 (Högbladh, 2011). Similar studies have been conducted, but no other study has used this
same dataset to extensively scrutinize the relationship between power-sharing and other
agreement provisions, including amnesty, reconciliation, elections, DDR, third party presence,
and peacekeeping. Krienbuehl (2010) uses an earlier version of the dataset, although there are
key differences between our models. First, Krienbuehl examines power-sharing following
interstate and intrastate wars, whereas my analysis looks specifically at the role of power-sharing
following civil wars because these wars are characteristically different than interstate wars.
Second, Krienbuehl bases his research on Lijphart’s original definition of power-sharing,
whereas my research utilizes the more broadly-defined and commonly used definition shaped by
Hartzell and Hoddie (2003). Finally, his research is qualitative in nature, in which he uses a casestudy approach to examine eight peace agreements that occurred from 1989 to 2005; whereas my
study is quantitative in nature. Melander (2009) also uses an earlier version of the dataset, but his
study focuses on the role of amnesty and political gender inequality in preventing violence, not
power-sharing. In short, no other study has used this version of the dataset or this model.
In this dataset, the peace agreement is the unit of analysis, meaning that there may be
multiple agreements for one conflict and therefore each agreement is analyzed separately.
Moreover, while a civil war may involve a complex set of actors, including a government and
two or more rebel groups, the dataset treats each agreement as if it were signed by a dyad which
includes only the “primary warring party on side A” and the “primary warring party on side B”
(Högbladh, 2011). These differentiations allow us to ascertain which accords contributed to a
cessation of fighting between which groups. This dataset also includes information on the name
of the warring parties, the name of the accord, the date the deal was signed, information on
provisions, and whether conflict reconvened after a period of time.
27 Earlier studies have only researched agreements signed by parties involved in large civil
wars with at least 1000 battle deaths per year, while this dataset includes wars with at least 25
battle deaths per year between a government and a rebel group. There are several justifications
for using a lower threshold. First, low intensity conflicts are common and they entail significant
human and financial consequences (domestically and globally) (Melander, 2009). Second,
smaller wars have the potential to turn into more destructive large-scale wars. Hence,
understanding how to end lower intensity wars is arguably just as critical to security as ending
larger conflicts. Moreover, excluding smaller-scale wars could mean that I would risk
underestimating the extent to which power-sharing fosters peace if “power-sharing provisions
are more likely to be included in peace agreements that are negotiated in very large wars” and if
“peace agreements in very large wars are generally more likely to fail” (Melander, 2009).
A peace deal may be considered a failure if civil war reignites at any point following the
signing of the agreement (Hartzell and Hoddie, 2003). However, research studies must
necessarily define the durability of peace more narrowly due to data limitations as well as the
desire to include more recent wars to examine modern trends. Moreover, many argue that if an
agreement brings peace for at least a few years, then it was more or less successful in bringing
conflict to an end. Thus, most researchers today opt to operationalize durability as a period of
short-term peace between signatories. Binningsbø (2005) operationalizes durability as a one-year
period of peace. Melander (2009) uses a two-year period because some evidence suggests that
peace is most fragile during the first two years after a peace signing. However, the emerging
consensus within the literature is that a five-year period of peace provides the most accurate
account of peace durability, because once nations have remained at peace for five years, they are
highly unlikely to return to war (Hartzell and Hoddie, 2003; Walter, 2002; Jarstad and Nilsson,
28 2008; Krienbuehl, 2012). Hence, to assess whether the agreement led to durable peace between
the dyad, this paper uses a binary dummy that assesses whether “violence with the same parties
restarted within 5 years” following the signing of the peace deal as its dependent variable (coded
by UCDP as DyVi05) (Högbladh, 2012).
While examining the agreement itself is beneficial for assessing the effects of the
agreement, a clear limitation of this method is that the number of observations is limited.
Additionally, I excluded twenty peace accords signed during wars involving interstate violence. I
also excluded two civil wars that occurred after 2008, in order to use the 5 year threshold for
peace durability. Therefore, I’ve included a total of only 195 accords signed between parties
involved in one of 53 civil wars.
UCDP variables that most closely resembled the contemporary definitions of powersharing are used to operationalize these provisions. The independent variables of interest are a
political power-sharing dummy variable indicating whether “the agreement provided for the
integration of rebels into the government or civil service (intgov), a military power-sharing
dummy that reflects whether “the agreement provided for the creation of a new national army or
the integration of rebels into the army” (intarmy), and a territorial power-sharing dummy to
define whether “the agreement granted the disputed region power-sharing in the local
government” (shaloc) (Högbladh, 2012). Due to limited data and access to information, I was
unable to obtain data regarding economic power-sharing.
The dataset includes detailed information regarding other provisions of the settlement,
allowing me to assess their impact on peace in relation to power-sharing. For example, the
UCDP amnesty dummy indicates whether “the agreement provided for an amnesty” (amn), the
reconciliation dummy defines whether “the agreement included the concept of National
29 Reconciliation” (recon), and the election dummy is operationalized as “the agreement provided
for elections or stipulated electoral reforms” (elections). Another dummy expresses whether “the
agreement included provisions for the disarmament of the warring parties” that is “coded as yes
even if the disarmament only concerns one of the warring parties” (ddr) (Högbladh, 2011). I
used the UCDP third party and peacekeeping dummies to create two new variables. Third_pko is
a dummy that indicates whether the agreement stipulated for the deployment of a peacekeeping
operation and whether a third party monitored the peace process.” Third_nopko is a dummy that
expresses whether a third party participated in the peace process without a peacekeeping
operation (the baseline category is that a third party was not involved in the peace process). This
dataset identifies whether an agreement includes both parties of the dyad or whether one of the
parties to the conflict is excluded (inclusive). The dataset also identifies the type of peace
agreement. Full (or comprehensive) agreements are those in which the dyad agrees to address the
entire range of issues underlying the conflict (coded as 1), partial agreements are those in which
at least one of the parties agrees to address some of the underlying conflict issues (coded as 2),
and a peace process agreement coded as 3 in which “one or more dyad agrees to initiate a
process that aims to settle” all or part of the dispute at a later point (pa_type).
To examine the impact of conflict dynamics, or characteristics of the previous war, I
include a dummy from the UCDP/PRIO Armed Conflict Dataset v.4-2012 that denotes the
cumulative intensity of the conflict coded as 1 if the “conflict since the onset has exceeded 1,000
battle-related deaths,” 0 if otherwise (cumint) (Themnér, Lotta & Peter Wallensteen, 2012). By
taking the difference between the date the conflict started (d_1_arm) from the UCDP Database
on Categorical Variables and the date that the agreement was signed (pa_date), I created a
continuous variable that gives the duration of the conflict in years (dur_yr) (Högbladh, 2011).
30 To account for a nation’s initial conditions, I’ve included Fearon, Kasara and Laitin’s
(2007) index for ethnic fractionalization that ranges from 0 to 1, where 0 is highly fragmented,
0.5 is generally “bipolar” or divided between two major ethnic groups, and 1 represents a
sizeable ethnic majority and fairly homogenous population (ethfrac). I’ve included Polity IV
(p4v2010) scores that express regime type based on the regime’s “qualities of executive
recruitment, constraints on executive authority, and political competition.” Polity scores range
from -10 to -6 (fully institutionalized autocracies), -5 to +5 (anocracies or mixed regimes), and
+6 to +10 (fully institutionalized democracies) (Marshall, Jaggers, and Gurr, 2011). Using data
from the United Nations (2013), I added a continuous variable for GDP per capita (GDPpc). To
capture dependence upon natural resources, I’ve created dummies that indicate whether a nation
is a diamond exporter (diamond) or an oil or gas exporter (oilgas) based on data from the Center
for the Study of Civil War’s International Peace Research Institute Oslo (Gilmore et. al., 2007;
Thieme et. al., 2007). I analyze the data using logistic regression analysis with robust standard
errors.
Results
I run three different logit models summarized in three tables in which I regress the
existence of political, military, and territorial power-sharing provisions to assess their impact on
the durability of peace, while controlling for other peace agreement provisions (Model 2, all
tables), conflict dynamics (Model 3, all tables), initial country conditions (Model 4, all tables),
and the fully specified model (Model 5, all tables). First, Figure A gives the total number of
agreements in the dataset that include political, military, or territorial power-sharing and gives
the number and percent of those agreements that resulted in either renewed war or durable peace.
As shown below, 75% of the political power-sharing agreements, 64% of the military power-
31 sharing agreements, and 78% of the territorial power-sharing agreements resulted in at least five
years of peace. Yet, logistic regression reveals more nuanced results, in which it becomes clear
that a number of other factors are also responsible for whether peace endured.
FIGURE A: Power-Sharing Agreements that Led to Durable Peace
(Data source: UCDP Dataset - Högbladh, 2011)
Type of Power-Sharing
Political Power-Sharing
Military Power-Sharing
Territorial Power-Sharing
Renewed War
8
24
2
Durable Peace
24
43
7
Total
32
67
9
Percent that resulted in peace
75%
64%
78%
As indicated in Table 1 (Model 1), logistic regression reveals that political power-sharing
provisions significantly reduce the probability that violence will restart among the dyad within
five years of signing a peace agreement. In fact, the chance for further violence decreases by a
factor of 0.44. When controlling for provisions commonly included in agreements, the overall
model remains significant (Prob > chi2 =0.0016), but interesting results arise (Model 2, Table 1).
Specifically, the influence of political power-sharing on war outcomes becomes insignificant.
This finding indicates that Model 1 suffered from omitted variable bias and overestimated the
peace-promoting impact of political power-sharing. Of all the provisions, only elections
significantly decrease the probability of renewed war (by a factor of 0.467). Contrary to several
prominent studies, agreements that were overseen by third parties (with and without a
peacekeeping operation) have a significantly higher probability of breaking down compared to
settlements without a third party. Finally, peace process agreements and partial agreements that
failed to address the full range of key underlying issues of the conflict were more likely to lead to
renewed war compared to comprehensive agreements (significant at the 95% confidence level).
32 When controlling for conflict dynamics, the overall model remains significant and
political power-sharing significantly reduces the odds of further violence (Model 3, Table 1). Yet
conflict duration and intensity do not significantly predict peace outcomes. Model 4 (Table 1)
controls for initial country conditions including structural and cultural factors and is significant
at the 95% level (Prob > chi2 value = 0.02). Model 4 reveals that low ethnic fractionalization and
fully institutionalized democracies significantly cut the chances for further warfare (at the 90 and
99 percent confidence level respectively). In contrast, GDP per capita and exportation of
diamonds, oil or gas do not considerably impact peace breakdown. However, when accounting
for a nation’s initial conditions, political power-sharing reduces the chances that the groups will
revert back into war (95% confidence level). Intriguingly, in the fully specified model (Model 5),
which includes agreement provisions, political power-sharing becomes significant again. I
cautiously interpret these results to mean that other arrangements within settlements are partially
predicting the impact of political power-sharing, which would explain why power-sharing is
insignificant when controlling for these variables. Moreover, the fact that the fully specified
model includes fewer observations than the peace agreement model (Model 2) may help explain
why political power-sharing is suddenly significant in the fully specified model even with the
inclusion of other agreement variables. The fully specified model also reveals that the presence
of a third party, the presence of a third party combined with a peacekeeping operation, and
diamond exportation each significantly increase the probability of renewed war. Finally, peace
process and partial agreements are more likely to collapse compared to comprehensive
agreements.
33 TABLE 1. Logistic Regression Results for Influence of Political Power-Sharing on Peace Durability
Variable Category
Independent Variable
Model 1
Model 2
Model 3
Model 4
Model 5
Power-Sharing
Political Power-Sharing
-0.840*
(-0.056)
-1.324***
(-0.008)
-1.176**
(0.013)
Peace Agreement
Amnesty
-0.574
(-0.268)
-0.496
(0.231)
-0.136
(0.73)
-0.761*
(0.06)
-0.543
(0.14)
2.416***
(0)
1.336***
(0.006)
0.346
(0.335)
0.647**
(0.016)
-1.052*
(0.094)
-0.721
(0.19)
-0.215
(0.677)
-1.002
(0.113)
-0.217
(0.706)
2.733**
(0.011)
1.107*
(0.094)
-0.361
(0.475)
1.300***
(0.005)
-0.002
(0.891)
0.088
(0.92)
-0.104
(0.901)
0
(0.644)
-0.061
(0.238)
1.511*
(0.067)
-0.227
(0.629)
-2.289
(0.237)
35.624
125
Reconciliation
Elections
DDR
Third Party and Peacekeeping
Third Party (No Peacekeeping)
Inclusiveness
Type of Peace Agreement
Conflict Dynamics
Conflict Duration (Years)
-0.011
(0.377)
0.546
(0.302)
Cumulative Intensity
Initial Conditions
Ethnic Fractionalization
-0.259
(-0.102)
-2.724***
(-0.008)
-0.14
(-0.777)
-1.122*
(0.083)
0
(0.645)
-0.102***
(0.002)
0.21
(0.617)
0.221
(0.505)
0.457
(-0.295)
3.659
195
26.875
194
7.567
129
14.033
188
GDP Per Capita
POLITY Score
Diamond Exporter
Oil or Gas Exporter
_constant
chi2
Observations
Notes: Dependent variable is a binary dummy that assesses whether violence with the same parties restarted within 5 years following
the signing of a peace agreement that contained political power-sharing. P-values are in parenthesis. *, **, and *** indicate
significance on a 90%, 95%, and 99% confidence level.
34 Table 2 assesses the impact of military power-sharing. Without controlling for other
factors, military power-sharing is not a significant predictor of relapse (Model 1, Table 2). It
continues to be insignificant when controlling for other provisions (Model 2, Table 2). Rather,
the terms that significantly decrease the odds of relapse are amnesty, elections, and DDR. On the
other hand, third party oversight, with and without a peacekeeping operation, significantly raise
the odds of renewed war. Additionally, partial agreements are associated with a higher risk of
conflict compared to comprehensive agreements. Model 3 (Table 2) indicates that conflict
duration and intensity are not significant predictors of peace outcomes, but when controlling for
these conflict dynamics, I find that military power-sharing significantly decreases the probability
of conflict relapse. However, it is important to note that Model 3 is insignificant as a whole
indicating the presence of omitted variables. When controlling for country conditions, military
power-sharing again significantly reduces the risk of future war (Model 4, Table 2). Yet this may
be attributed to the reduced number of observations in this model. The only initial condition that
significantly predicts peace outcomes is regime type; democracies are less likely to resume war
than authoritarian or mixed regimes. Lastly, in the fully specified model, military power-sharing
is insignificant (Model 5, Table 2). Instead, third parties paired with a peacekeeping operation,
diamond exportation, and partial agreements significantly increase the probability of war.
Table 3 provides some striking results. In particular, territorial power-sharing is
insignificant across all five models, indicating that providing rebels with the opportunity to share
power in local governments is insufficient for ending conflict. Rather, when controlling for the
terms of the settlement in Model 2, only elections significantly reduce chances for war whereas
third parties, with and without peacekeepers, and partial agreements increase the odds of conflict.
35 TABLE 2. Logistic Regression Results for Influence of Military Power-Sharing on Peace Durability
Variable Category
Independent Variable
Model 1
Model 2
Model 3
Model 4
Model 5
Power-Sharing
Military Power-Sharing
-0.3
(0.336)
-0.803**
(0.031)
-0.633*
(0.074)
Peace Agreement
Amnesty
0.491
(0.247)
-0.685*
(0.098)
-0.31
(0.425)
-0.836**
(0.035)
-0.633*
(0.092)
2.375***
(0)
1.314***
(0.009)
0.097
(0.794)
0.713***
(0.01)
-0.907
(0.174)
-0.769
(0.153)
-0.438
(0.389)
-1.021
(0.13)
0.106
(0.863)
2.467**
(0.014)
1.069
(0.112)
-0.342
(0.496)
1.367***
(0.002)
0.006
(0.687)
0.165
(0.833)
0.162
(0.844)
0
(0.703)
-0.092
(0.117)
1.598**
(0.031)
-0.229
(0.638)
-2.59
(0.185)
37.883
125
Reconciliation
Elections
DDR
Third Party and Peacekeeping
Third Party (No Peacekeeping)
Inclusiveness
Type of Peace Agreement
Conflict Dynamics
Conflict Duration (Years)
-0.002
(0.844)
0.595
(0.227)
Cumulative Intensity
Initial Conditions
Ethnic Fractionalization
_constant
-0.283
(0.114)
-2.576**
(0.011)
-0.236
(0.611)
-0.751
(0.24)
0
(0.522)
-0.105***
(0.002)
0.185
(0.648)
0.216
(0.51)
0.269
(0.532)
chi2
Observations
0.925
195
27.876
194
5.598
129
10.595
188
GDP Per Capita
POLITY Score
Diamond Exporter
Oil or Gas Exporter
Notes: Dependent variable is a binary dummy that assesses whether violence with the same parties restarted within 5 years following
the signing of a peace agreement that contained military power-sharing. P-values are in parenthesis. *, **, and *** indicate
significance on a 90%, 95%, and 99% confidence level.
36 Model 3 (Table 3) indicates that conflict dynamics are not strong predictors of falling
back into war. Interestingly, as Model 4 (Table 3) indicates, higher democracy scores are the
only initial condition that significantly decreases the chances of future violence (significant at the
99% confidence level). Finally, in the fully specified model (Model 5, Table 3), amnesty
decreases the odds of renewed war (p<0.1) whereas third parties paired with peacekeeping
forces, partial agreements, and diamond exportation significantly increase the odds of war.
Logistic regression provides important insight into the impact of power-sharing
agreements; however, there are several caveats. First, a word of caution must be stated when
interpreting these results because most accords in the sample did not include political, military,
or territorial power-sharing; hence settlements without these provisions were overrepresented.
Second, the fully specified model and model 3 in all three tables included fewer observations
(125 and 129 observations respectively). Third, not all of the assumptions of logit are fulfilled
because the observations were not randomly sampled. However, as a limited number of civil
wars have ended via a peace settlement, it was necessary for UCDP to hand select the cases.
Moreover, significant imperfect multi-collinearity, which would dampen the significance of
results, did not manifest in simple pair-wise correlations tests1. Another limitation is that powersharing arrangements were defined quite broadly, and it may be important to examine terms of
power-sharing arrangements more narrowly. For example, to examine the impact of grand
coalitions, proportional representation, and mutual veto power separately to discover which
political arrangements are most critical to peace. Finally, some models were not highly
significant, indicating the presence of omitted variables that better predict peace outcomes.
Nevertheless, these problems are common in studies that examine peace agreements.
1 Pairwise correlation matrices indicate that the independent variables are not highly correlated and variance inflation factors (VIF scores)
conducted on the fully specified models indicate that there is no significant multi-collinearity within the models. 37 TABLE 3: Logistic Regression Results for Influence of Territorial Power-Sharing on Peace Durability
Variable Category
Independent Variable
Model 1
Model 2
Model 3
Model 4
Model 5
Power-Sharing
Territorial Power-Sharing
-0.905
(0.268)
-1.445
(0.201)
-0.962
(0.244)
Peace Agreement
Amnesty
-0.117
(0.903)
-0.595
(0.143)
-0.262
(0.501)
-0.783*
(0.052)
-0.527
(0.149)
2.337***
(0)
1.284***
(0.009)
0.248
(0.464)
0.658**
(0.014)
-0.572
(0.734)
-1.006*
(0.051)
-0.406
(0.443)
-0.968
(0.131)
-0.159
(0.777)
2.360**
(0.022)
0.996
(0.148)
-0.544
(0.257)
1.379***
(0.003)
0.003
(0.859)
0.029
(0.971)
0.099
(0.907)
0
(0.709)
-0.064
(0.206)
1.520*
(0.064)
-0.18
(0.702)
-2.306
(0.247)
37.312
125
Reconciliation
Elections
DDR
Third Party and Peacekeeping
Third Party (No Peacekeeping)
Inclusiveness
Type of Peace Agreement
Conflict Dynamics
Conflict Duration (Years)
-0.004
(0.705)
0.466
(0.355)
Cumulative Intensity
Initial Conditions
Ethnic Fractionalization
-0.348**
(0.02)
-2.582**
(0.011)
-0.366
(0.436)
-0.843
(0.177)
0
(0.504)
-0.087***
(0.008)
0.219
(0.589)
0.3
(0.359)
0.069
(0.866)
1.226
195
26.066
194
2.433
129
8.738
188
GDP Per Capita
POLITY Score
Diamond Exporter
Oil or Gas Exporter
_constant
chi2
Observations
Notes: Dependent variable is a binary dummy that assesses whether violence with the same parties restarted within 5 years following
the signing of a peace agreement that contained territorial power-sharing. P-values are in parenthesis. *, **, and *** indicate
significance on a 90%, 95%, and 99% confidence level.
38 Discussion and Policy Implications
Several findings may be gleaned from this research that have important policy
implications. Intensity and duration of the conflict were not significant predictors of peace
outcomes in the 53 civil wars included in this dataset. However, this is not to say that these
variables are unimportant, as several other studies have found evidence that these characteristics
significantly impact the probability of renewed war. Likewise, many of the initial conditions of
the country did not significantly impact the probability of peace, although the results demonstrate
that conflict negotiators are bound to experience greater difficulty establishing peace in nations
that export diamonds, score low on democracy, and that have higher ethnic fractionalization. Yet
as several other researchers have found, the structural and cultural characteristics of the country
prior to the war often influence whether war recurs in one way or another (Collier, 2007).
Despite the difficulty of establishing peace, this research indicates that peace is something that
can be facilitated through the institutional arrangements included in negotiated settlements.
Additionally, this research provides further evidence that comprehensive agreements (that
include multiple arrangements to address a greater range of the parties’ concerns) are more likely
to lead to durable peace than partial agreements or merely initiating a peace process. In other
words, instituting multiple provisions will provide greater chances for lasting peace.
A central discovery of this research is that power-sharing decreases the probability that
war will restart among the same parties within five years of signing a peace accord. Yet, some
forms of power-sharing promote peace better than others. For example, political power-sharing,
or integrating rebels into the government or civil service, proved to be more effective for ending
war compared to military and territorial power-sharing. In other words, when former adversaries
share government power via mechanisms such as grand coalitions, proportional representation,
39 and a system of mutual veto-power, they are less likely to initiate further warfare. However,
when controlling for other factors it becomes clear that the significance of shared political power
is contingent upon model specification. Specifically, political power-sharing is successful at
diminishing the risk of conflict without controlling for other factors and when accounting for
conflict intensity, conflict duration, GDP per capita, ethnic fractionalization, regime type, and
natural resources exportation. Thus, political power-sharing is significant across all of the models
except Model 2, which controls for other agreement provisions. Military power-sharing also
significantly reduces the probability of conflict relapse, but it is not as effective as political
power-sharing. In particular, integrating rebels into the national army only mitigates conflict
when accounting for initial conditions of the nation and the dynamics of the previous war.
Perhaps when state soldiers and rebel soldiers are integrated into the same army, they are better
able to develop a common identity and move beyond their differences. It is critical to note that
territorial arrangements, which grant rebels political representation in the local government,
appear to be ineffectual in reducing the risk of conflict. Perhaps rebels are discontented when
they merely gain more control over local issues rather than matters of the state. Certainly an
important area for future research is to examine the nature of the relationships between powersharing and the independent variables in this study, such as the conflict dynamics and initial
characteristics of the nation. For example, is it that political power-sharing is most beneficial for
reducing conflict risk in small scale wars or large scale wars? Qualitative case studies may be
more beneficial for uncovering these issues in greater detail. While further research is needed to
assess why the literature provides such conflicting results, my research adds to the body of
literature that finds that political and military power-sharing increase the chances for peace.
40 Thus, a practical implication of my research is that when mediators are designing settlements,
they should prioritize political and military power-sharing over territorial power-sharing.
Remarkably, another principal finding is that when taking into account the other terms
included in the peace settlement across all three tables, all forms of power-sharing (political,
military, and territorial) are insignificant predictors of peace. This suggests that something else
within the peace agreement is driving peace outcomes and therefore negotiators should prioritize
these provisions over power-sharing. The provisions most effective for diminishing the
probability of war are DDR, elections, and amnesty.
As discussed in the literature, disarmament, demobilization, and reintegration of soldiers
into civil society is a critical first step to signal commitment to peace (Stedman, 2001). It is
necessary to note, however, that it is often only the rebels who give up their arms after a civil
war because the state is typically not expected to give up its military resources. Rebels
commonly experience greater security threats during this period and it is therefore recommended
that a third party oversee demobilization and disarmament, as well as promise to intervene
should either party take advantage of the vulnerability of their opponent.
Additionally, amnesty proved to facilitate war termination in several cases. Likely
because reducing fears of prosecution encourages groups to disarm and demobilize (Melander,
2009). There are valid human rights concerns and disadvantages to granting amnesty to
combatants and/or leaders of the war, however, the reality may be that it is an effective and
important tool for bringing conflict to a close. Fewer studies have examined the conditions under
which amnesty facilitates peace in relation to power-sharing, and given its controversial nature it
also warrants further scrutiny.
41 Two other key findings are that democracies are less likely to return to war compared to
mixed or authoritarian regimes and that holding elections decreases the chance of relapse. Again,
power-sharing was not a predictor of peace when controlling for other settlement terms, such as
elections, indicating that elections are more important for establishing long-term peace. This
finding lends support to the idea that if power-sharing is to be used, it should be used only as a
short-term solution. Instead, long-term peace requires that efforts be made to help the country
transition toward a more democratic form of governance via elections.
Nevertheless, elections are not a panacea. It still holds true that holding elections too
early threatens the security of the nation if it lacks the institutional capacity to run elections
smoothly or if losing groups decide to return to war. Several studies provide evidence that
ensuring democratic and peaceful elections requires substantial political, judicial, and electoral
system reforms to occur first (Norris, 2008; Sisk, 2003). For example, reforms that sufficiently
check the power of the president (such as strict term limits) are important for ensuring that the
nation will be able to hold regular elections, which helps losing groups believe that they have a
chance to compete fairly in the next election. In other words, it is unlikely that the nation will be
ready to hold elections immediately after the war. Therefore, negotiators may need to use
political or military power-sharing or some other transitional government to get the groups to
stop fighting and then assist the parties in holding democratic elections once the nation has
undergone significant reform and is stable enough to support such a system.
Lastly, all three tables demonstrate that the presence of a third party intervention both
with and without a peacekeeping operation significantly increases the chances that the conflict
dyad will revert back into war. Yet, it is important to be cautious when interpreting these results
as they contradict many previous studies which indicate that competent and engaged third party
42 enforcers are largely successful at keeping the peace (Mattes and Savun, 2010; Kieh Jr., 2011). It
is entirely plausible, however, that several of the third parties involved in these 53 civil wars
undermined peace due to a lack of capacity or bias. Third parties are often willing to observe the
negotiating process and verify information, but far fewer are willing to take on the risks of
enforcing the peace. Additionally, third parties may show favoritism or may fail to report
attempts to spoil the peace out of fear of appearing biased and experiencing retaliation (Stedman,
2001). However, the deployment of third party interventions is not random and therefore the
relationship could be spurious. As discussed in the literature, peacekeepers are often sent to
nations “where they are most needed, when peace would otherwise be difficult to keep” which
would cause us to “underestimate the effectiveness of peacekeeping” and third party involvement
(Fortna, 2004). Hence, an important area for further research is to uncover why third party
intervention is associated with higher risk of war within this dataset.
Conclusion
Civil conflicts are raging in Syria, Iraq, Afghanistan, Colombia, Somalia, Mali, and in
many other regions of the world. These wars are costly and the problems associated with these
wars are spilling over into other nations each day. Certainly, one of the most pressing global
needs is to develop more refined policies for ending civil war and unrest. Yet there are many
challenges to developing and implementing such policies. Internal armed conflict is particularly
persistent and getting warring parties to agree to a bargained solution in the face of fear and
uncertainty is arduous. However, this research demonstrates that institutional arrangements
matter and that political and military power-sharing help end war when taking into account the
structural and cultural conditions of the nation as well as the characteristics of the previous war.
These arrangements facilitate elite support of the peace process by leveling the playing field and
43 helping parties credibly signal their commitment to peace. However, it is important to keep in
mind that power-sharing, as some of the literature suggests, may aggravate conflict or have no
impact on peace.
Additionally, this research indicates that other settlement provisions play a greater role in
bringing conflict to an end than power-sharing and therefore negotiators should prioritize
comprehensive agreements that provide for disarmament, demobilization, and reintegration of
soldiers, amnesty and elections. Peace is more likely to endure when nations transition away
from power-sharing and toward a more democratic form of governance via free and fair
elections. There are no simple solutions for ending civil war as every country poses a unique set
of circumstances that impact the prospects of war termination. Thus, when crafting military
and/or political power-sharing agreements, it is necessary to consider the conditions of the
country and of the conflict to decide which power-sharing provisions are most appropriate and
whether such arrangements are necessary.
44 References
Binningsbø, Helga Malmin. (2006). “Consociational Democracy and Postconflict Peace. Will PowerSharing Institutions Increase the Probability of Lasting Peace after Civil War?” Paper presented
at the annual meeting of the International Studies Association, Town & Country Resort and
Convention Center, San Diego, California, USA. Retrieved from
http://www.allacademic.com/meta/p98892_index.html
Blattman, Chris and Edward Miguel. (2010). Civil War. Journal of Economic Literature, 48(2): 3-57. doi:
10.1257/jel.48.1.3
Brosché, Johan. (2009). Sharing Power – Enabling peace?: Evaluating Sudan’s Comprehensive Peace
Agreement 2005. Uppsala University, Department of Peace and Conflict Research. Uppsala,
Sweden.
Collier, Paul. (2007). The Bottom Billion. Oxford: Oxford University Press.
Demmers, Jolle. (2012). Theories of Violent Conflict: An Introduction. Taylor and Francis.
Derouen, Karl, Jr., Jenna Lea, and Peter Wallensteen. (2009). The Duration of Civil War Peace
Agreements. Conflict Management and Peace Science, 26(4): 367-387.
doi: 10.1177/0738894209106481
Fearon, James D. (1995). Rationalist Explanations for War. International Organization, 49(3): 379-414.
Retrieved from http://www.jstor.org/stable/2706903
Fearon, James D. (2003). Ethnic and Cultural Diversity by Country. Journal of Economic Growth, 8(2):
195-222. Retrieved from http://www.jstor.org/stable/40215943
Fearon, James D., Kimuli Kasara and David Laitin (2007). Ethnic Minority Rule and Civil War Onset
Replication Data [Data file]. Retrieved from http://www.stanford.edu/~jfearon/.
Gates, Scott and Kaare Strom. (2007). Power-sharing, Agency and Civil Conflict. Center for the Study of
Civil War, Oslo.
Gilmore, Elisabeth, Päivi Lujala, Nils Petter Gleditsch & Jan Ketil Rød. (2005). The Diamond Dataset
Codebook. Center for the Study of Civil War, PRIO.
Hartzell, Caroline and Matthew Hoddie (2003), Institutionalizing Peace: Power Sharing and Post-Civil
War Conflict Management. American Journal of Political Science, 47(2): 318-332. doi:
10.1111/1540-5907.00022
Hartzell, Caroline and Matthew Hoddie (2007). Crafting Peace: Power-Sharing Institutions and the
Negotiated Settlement of Civil Wars. University Park, PA: The Pennsylvania State University
Press. Retrieved from http://muse.jhu.edu/
Högbladh, Stina. (2011). Peace agreements 1975-2011 - Updating the UCDP Peace Agreement dataset
[Data file]. Pettersson Therése & Lotta Themnér (eds.), 2012, States in Armed Conflict 2011,
Uppsala University: Department of Peace and Conflict Research Report 99 Retrieved from
http://www.pcr.uu.se/research/ucdp/datasets/ucdp_peace_agreement_dataset/
45 Högbladh, Stina. (2012). UCDP Peace Agreement Dataset Codebook, version 2.0. Uppsala: Uppsala
Conflict Data Program.
Horowitz, Jeremy. (2009). “Power-Sharing in Kenya”. Prepared for the Workshop on Political Inclusion
in Africa, American University on April 24-25, 2009. University of California, San Diego.
Human Rights Watch. (2012). The Special Court for Sierra Leone’s Landmark Judgment in the Trial of
Charles Taylor. Retrieved from http://www.hrw.org/topic/international-justice/charles-taylor
Jarstad, Anna K. (2009). The Prevalence of Power-Sharing: Exploring the Patterns of Post-Election
Peace. Africa Spectrum. 44(3): 41-62. Retrieved from http://hup.sub.unihamburg.de/giga/afsp/article/view/191
Jarstad, Anna K., and Desireé Nilsson. (2008). From Words to Deeds: The Implementation of PowerSharing Pacts in Peace Accords. Conflict Management and Peace Science, 25(3): 206-223.
doi: 10.1080/07388940802218945
Kieh Jr., George Klay. (2011). Peace Agreements and the Termination of Civil Wars: Lessons from
Liberia. African Journal on Conflict Resolution, 11(3)
Krienbuehl, Lukas. (2010). Peace with Power-Sharing: Under Which Conditions? Swiss Peace
Foundation, Working Paper. Available at:
http://www.swisspeace.ch/countries/sudan/research.html
Lake, David A. (2010). Two Cheers for Bargaining Theory: Assessing Rationalist Explanations of the
Iraq War. International Security, 35(3): 7-52. The MIT Press. Retrieved from http://muse.jhu.edu/
Lijphart, Arend. (1969). Consociational Democracy. World Politics. 21(2): 207-225.
Lijphart, Arend. (1993). Consociational Democracy, in: Joel Krieger (ed.), The Oxford Companion to
Politics of the World, New York, Oxford: Oxford University Press.
Lijphart, Arend. (1999). Patterns of Democracy. Government Forms and Performances in Thirty-Six
Countries. New Haven and London: Yale University Press.
Maiese, Michelle. (2003). Addressing Injustice. Beyond Intractability. Eds. Guy Burgess and Heidi
Burgess. Conflict Information Consortium, University of Colorado, Boulder. Retrieved from
http://www.beyondintractability.org/bi-essay/address-injustice
Malek, Cate. (2013). Reconciliation. Beyond Intractability. Conflict Information Consortium, University
of Colorado, Boulder. Retrieved from
http://www.beyondintractability.org/coreknowledge/reconciliation
Marshall, Monty G., Keith Jaggers, & Ted Robert Gurr. (2011). Polity IV Project: Political Regime
Characteristics and Transitions, 1800-2010. Dataset User’s Manual. University of Maryland:
Integrated Network for Societal Conflict Research Program (INSCR) Program and Center for
International Development and Conflict management (CIDCM). Retrieved from
http://www.systemicpeace.org/inscr/p4manualv2010.pdf
46 Mason, David, T., Gurses, M., Brandt, P. T. and Michael Quinn, J. (2011), When Civil Wars Recur:
Conditions for Durable Peace after Civil Wars. International Studies Perspectives, 12: 171–189.
doi: 10.1111/j.1528-3585.2011.00426.x
Mattes, M. and Savun, B. (2009), Fostering Peace After Civil War: Commitment Problems and
Agreement Design. International Studies Quarterly. 53: 737–759. doi: 10.1111/j.14682478.2009.00554.x
Mattes, M. and Savun, B. (2010), Information, Agreement Design, and the Durability of Civil War
Settlements. American Journal of Political Science. 54(2): 511–524. Retrieved from
http://www.jstor.org/stable/25652220
Mehler, A. ( 2009 ). Introduction: Power-Sharing in Africa. Africa Spectrum, 44(3): 2-10. Retrieved
from http://hup.sub.uni-hamburg.de/giga/afsp/article/view/189
Melander, Erik. 2009. Justice or Peace? A Statistical Study of the Relationship between Amnesties and
Durable Peace. Lund University, JAD-PbP Working Paper Series (4).
Mukherjee, B. (2006), Why Political Power-Sharing Agreements Lead to Enduring Peaceful Resolution
of Some Civil Wars, But Not Others?. International Studies Quarterly. 50: 479–504.
doi: 10.1111/j.1468-2478.2006.00410.x
Norris, Pippa. (2008). Driving Democracy: Do Power-sharing Institutions Work? New York: Cambridge
University Press.
Nye, Joseph S. Jr. (2011). The Future of Power. New York: Public Affairs.
Ouellet, Julian. (2004). Structural Components of Peace Agreements. Beyond Intractability. Eds. Guy
Burgess and Heidi Burgess. Conflict Information Consortium, University of Colorado, Boulder.
Retrieved from http://www.beyondintractability.org/bi-essay/substantive-provisions
Peace Accords Matrix. (2012). Frequently Asked Questions. Kroc Institute for International Peace
Studies, University of Notre Dame. Notre Dame, IN. Retrieved from
https://peaceaccords.nd.edu/faq
Polity IV Project. (2011). Polity IV: Regime Authority Characteristics and Transitions Datasets (Polity IV
Annual Time-Series 1800-2011 [Excel Time series data file]. Center for Systemic Peace.
Available at http://www.systemicpeace.org/inscr/inscr.htm
Rotberg, Robert I.; Thompson, Dennis (2010). Truth v. Justice: The Morality of Truth Commissions.
Retrieved from
http://osu.eblib.com.ezproxy.proxy.library.oregonstate.edu/patron/FullRecord.aspx?p=483525
Sisk, Timothy D. (2003). Power Sharing. Beyond Intractability. Eds. Guy Burgess and Heidi Burgess.
Conflict Information Consortium, University of Colorado, Boulder. Retrieved from
http://www.beyondintractability.org/bi-essay/power-sharing
Snyder, Jack, and Leslie Vinjamuri (2003). Trials and Errors: Principle and Pragmatism in Strategies of
International Justice. International Security, 28 (3):5-44.
Stedman, Stephen John. (1997). Spoiler Problems in Peace Process. International Security, 22(2): 5-53.
47 Stedman, Stephen John. (2001). Implementing Peace Agreements in Civil Wars: Lessons and
Recommendations for Policymakers. International Peace Academy and Center for International
Security and Cooperation. New York. Retrieved from
http://www.ipacademy.org/PDF_Reports/Pdf_Report_Implementing.pdf
Themnér, Lotta & Peter Wallensteen. (2012). Armed Conflict, 1946-2011. Journal of
Peace Research, 49(4): 565-575.
Thieme, Nadja; Päivi Lujala & Jan Ketil Rød. (2007). The Petroleum Dataset: Country Profiles [Data
File]. Center for the Study of Civil War, PRIO. Retrieved from
http://www.prio.no/Global/upload/CSCW/Data/Geographical/Country%20Profiles%202007.pdf
Tull, D. M. and Mehler, A. (2005). The Hidden Costs of Power-Sharing: Reproducing Insurgent Violence
in Africa. African Affairs, 104(416):. 375-398. doi: 10.1093/afraf/adi034
United Nations. (2013). National Accounts Main Aggregates Database (GDP and its breakdown at
current prices in US Dollars – All countries for all years – sorted alphabetically [Data file].
United Nations Statistics Division. Retrieved from http://unstats.un.org/unsd/snaama/dnllist.asp
Uppsala Conflict Data Program (UCDP). (2013). UCDP Conflict Encyclopedia [Data file]. Retrieved
from www.ucdp.uu.se/database
Uppsala Conflict Data Program (UCDP) . (2013). UCDP Database Categorical Variables 1989-2008
[Data file]. Uppsala University. Retrieved from www.ucdp.uu.se
VanderZee, Lenore, Laura Taylor, Alex Dukalskis, Alma Gottlieb-McHale, and Chris Sullivan. (2010).
Comprehensive Peace Agreement: How is it Defined and Why Does It Matter? Peace Accords
Matrix. Retrieved from www.peaceaccords.org
Wallensteen, Peter. (2001). The Growing Peace Research Agenda. Occasional Paper No. 21. Joan B.
Kroc Institute for International Peace Studies, San Diego, CA.
Walter, Barbara F. (2002). Committing to Peace: The Successful Settlement of Civil Wars. Princeton:
Princeton: University Press.
Walter, Barbara F. (2011). Conflict Relapse and the Sustainability of Post-Conflict Peace. Washington,
DC: World Bank. Retrieved from https://openknowledge.worldbank.org/handle/10986/9069
Werner, Suzanne and Amy Yuen (2005). Making and Keeping Peace. International Organization, 59(2):
261-292. doi:10.1017/S0020818305050095.
Yawanarajah, Nita and Julian Ouellet. (2003). Peace Agreements. Beyond Intractability. Eds. Guy
Burgess and Heidi Burgess. Conflict Information Consortium, University of Colorado, Boulder.
http://www.beyondintractability.org/bi-essay/structuring-peace-agree
48